The annual review is one of those tasks that feels important in theory and gets pushed back in practice. By the time December arrives, you're either in the middle of the holiday rush or winding down for a break — neither of which is ideal for sitting with your numbers and thinking hard about where you're going.
October is the right time to do this. You have almost a full year's data, the year isn't over yet (so you can still affect how it ends), and you have enough runway before January to actually act on what you find. Here's the process that makes it manageable.
The Half-Day Annual Review Structure
Gather Your Numbers First (30 Minutes)
Before you open Claude, pull the raw data. You need: total revenue by month, revenue by client or product line (if applicable), expenses by category, new clients or customers acquired, clients or customers lost, and any other metrics specific to your business (average order value, conversion rate, project turnaround time, etc.).
Don't analyse yet — just gather. Get it all into a single document or spreadsheet. The discipline of collecting everything before you start interpreting it prevents the common trap of spending two hours on the one number you can find easily and ignoring the others.
Monthly revenue (Jan–Oct), top 5 revenue sources, top 3 expenses, number of new clients, number of churned clients, average project value or order size, one or two operational metrics you track regularly. If you don't track all of these — that's useful information too. Note the gaps.
Use AI to Find the Patterns You're Too Close to See
Paste your numbers into Claude and ask it to identify patterns, anomalies, and questions worth investigating. The key instruction: tell it not to give you generic business advice, but to look at your specific data and tell you what it notices.
This step consistently surfaces things that owners miss when they're working inside the business every day. A revenue dip in March that you thought was a slow month might correlate exactly with a pricing change you made in February. A strong Q2 might be entirely explained by two clients who didn't return in Q3. The patterns are in the data — you just need the right questions to find them.
Here's my business data for 2026: [paste numbers]. My business is [describe in one sentence]. Please: (1) Identify the three most significant patterns or trends in this data. (2) Flag any anomalies that seem worth investigating. (3) List five questions I should be asking about this data that I probably haven't asked yet. Do not give generic advice — respond specifically to what you see in these numbers.
Run the "What Worked / What Didn't / What to Stop" Inventory
This is the qualitative part of the review. Write a brain dump — no more than 20 minutes — covering three questions: What were the three things that most contributed to growth or success this year? What were the three biggest problems, mistakes, or missed opportunities? What are you currently doing that's consuming time or money with little return?
Then paste this brain dump into Claude alongside your numbers and ask it to find the connections. Often the "what worked" maps directly to the data (a specific client type, a specific service, a specific channel), and the same is true for what didn't. Making those connections explicit is where the real insight lives.
Here's my year-end brain dump: [paste]. Here's my financial data: [paste summary]. Where do you see connections between what I've said worked and the numbers? Where do the problems I've identified show up in the data? What am I missing that the numbers might be pointing to?
Build Your 3-Priority Plan for Next Year
The output of a good annual review isn't a 20-page strategy document — it's three clear priorities for the year ahead. Businesses that can articulate three concrete priorities consistently outperform those with sprawling plans, because clarity drives execution.
Ask Claude to synthesise everything you've produced so far and propose three priorities for next year, ranked by likely impact. Give it the context to do this well: what you're trying to achieve overall, any constraints (budget, time, team size), and what you've learned from the review. Then challenge its output — push back on any priority that doesn't feel right and ask why it ranked above the alternatives.
Based on everything we've discussed — my 2026 data, what worked, what didn't, what to stop — propose three priorities for 2027, ranked by expected impact. For each: state the priority clearly, explain why you're suggesting it based on the specific evidence from my review, and suggest one concrete first action. My overall goal is [X]. My constraints are [time/budget/team].
Making the Output Stick
The review is only valuable if you act on it. The most common failure mode is producing a thoughtful analysis in October and then finding it again in August of the following year having done none of it.
Two practices help. First, turn your three priorities into specific 90-day milestones (what does success look like by end of January?). Second, build a recurring quarterly check-in: a 45-minute session in January, April, and July where you ask Claude to remind you of your priorities and compare your current state against them. This takes 5 minutes to schedule now and prevents the drift that kills most year-end planning.
The honest version: the value of this process isn't the document at the end. It's the thinking — the hour you spend actually looking at your numbers, naming what's working, and making deliberate choices about where to put your energy. AI makes this faster and sharper. It doesn't replace the hour. It makes it more productive.
If you start the review process in October, you have time to course-correct before the year ends, time to plan January properly, and time to arrive at the new year with intention rather than momentum from whatever happened to be on your plate in December.
Want Help Building Systems That Make Next Year Easier?
We help small businesses automate the operations that are slowing them down — so the year ahead looks different from the year that just finished. Get in touch to learn more.
Talk to us